Vector Autoregression Analysis and the Great Moderation

Most analyses of the U.S. Great Moderation have been based on VAR methods, and have consistently pointed toward good luck as the main explanation for the greater macroeconomic stability of recent years. Using data generated by a New-Keynesian model in which the only source of change is the move from passive to active monetary policy, we show that VARs may misinterpret good policy for good luck. In particular, we detect significant breaks in estimated VAR innovation variances, although in the data generating process the volatilities of the structural shocks are constant across policy regimes. Counterfactual simulations, structural and reduced-form, point toward the incorrect conclusion of good luck. Our results cast doubts on the existing notion that VAR evidence is inconsistent with the good policy explanation of the Great Moderation.